Merge pull request #244 from snshn/process-noscript

Process contents of NOSCRIPT tags
3 years ago · 6d629bfd4a
parent 0f55fb3c49 ae9d78a891
commit 6d629bfd4a
2 changed files with 79 additions and 2 deletions
--- a/src/html.rs
+++ b/src/html.rs
@ -1165,8 +1165,8 @@ pub fn walk_and_embed_assets(
                        // Empty inner content of STYLE tags
                        node.children.borrow_mut().clear();
                    } else {
-                        for node in node.children.borrow_mut().iter_mut() {
-                            if let NodeData::Text { ref contents } = node.data {
+                        for child_node in node.children.borrow_mut().iter_mut() {
+                            if let NodeData::Text { ref contents } = child_node.data {
                                let mut tendril = contents.borrow_mut();
                                let replacement = embed_css(
                                    cache,
@ -1402,6 +1402,42 @@ pub fn walk_and_embed_assets(
                        }
                    }
                }
+                "noscript" => {
+                    for child_node in node.children.borrow_mut().iter_mut() {
+                        match child_node.data {
+                            NodeData::Text { ref contents } => {
+                                // Get contents of the NOSCRIPT node
+                                let mut noscript_contents = contents.borrow_mut();
+                                // Parse contents of the NOSCRIPT node
+                                let noscript_contents_dom: RcDom = html_to_dom(&noscript_contents);
+                                // Embed assets within the NOSCRIPT node
+                                walk_and_embed_assets(
+                                    cache,
+                                    client,
+                                    &url,
+                                    &noscript_contents_dom.document,
+                                    &options,
+                                    depth,
+                                );
+                                // Get rid of original contents
+                                noscript_contents.clear();
+                                // Insert HTML containing embedded assets into the NOSCRIPT node
+                                if let Some(html) =
+                                    get_child_node_by_name(&noscript_contents_dom.document, "html")
+                                {
+                                    if let Some(body) = get_child_node_by_name(&html, "body") {
+                                        let mut buf: Vec<u8> = Vec::new();
+                                        serialize(&mut buf, &body, SerializeOpts::default())
+                                            .expect("Unable to serialize DOM into buffer");
+                                        let result = String::from_utf8(buf).unwrap();
+                                        noscript_contents.push_slice(&result);
+                                    }
+                                }
+                            }
+                            _ => {}
+                        }
+                    }
+                }
                _ => {}
            }

--- a/src/tests/html/walk_and_embed_assets.rs
+++ b/src/tests/html/walk_and_embed_assets.rs
@ -326,4 +326,45 @@ mod passing {
            </html>"
        );
    }
+
+    #[test]
+    fn processes_noscript_tags() {
+        let html = "<html>\
+            <body>\
+                <noscript>\
+                    <img src=\"image.png\" />\
+                </noscript>\
+            </body>\
+        </html>";
+        let dom = html::html_to_dom(&html);
+        let url = "http://localhost";
+        let cache = &mut HashMap::new();
+
+        let mut options = Options::default();
+        options.no_images = true;
+        options.silent = true;
+
+        let client = Client::new();
+
+        html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
+
+        let mut buf: Vec<u8> = Vec::new();
+        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+
+        assert_eq!(
+            buf.iter().map(|&c| c as char).collect::<String>(),
+            format!(
+                "<html>\
+                    <head>\
+                    </head>\
+                    <body>\
+                        <noscript>\
+                            <img src=\"{}\">\
+                        </noscript>\
+                    </body>\
+                </html>",
+                empty_image!(),
+            )
+        );
+    }
 }